Brain Hypometabolism of Glucose in Anorexia
Nervosa: Normalization after Weight Gain
Vrronique Delvenne, Serge Goldman, Viviane De Maertelaer, Yves Simon,
Andr6 Luxen, and Franqoise Lotstra
Using positron emission tomography and (18-F)-fluorodeoxyglucose, we studied cerebral
glucose metabolism in 10 anorectic girls within their underweight state and after weight gain.
Ten age- and sex-matched healthy volunteers were used as controls. Both groups were
scanned during rest, eyes closed and with low ambient noise. In absolute values, the
underweight anorectic patients, when compared to control subjects, showed a global (p =
0.002) and regional (p <- 0.001) hypometabolism of glucose which normalized with weight
gain. In relative values, no global difference could be assessed between underweight anorectic
patients and controls but a trend can, nevertheless, be observed toward parietal and superior
frontal cortex hypometabolism associated with a relative hypermetabolism in the caudate
nuclei and in the inferior frontal cortex. After weight gain, all regions normalized for absolute
and relative values, although a trend appears toward relative parietal hypometabolism and
inferior frontal cortex hypermetabolism in weight gain anorectic patients.
Absolute brain glucose hypometabolism might result from neuroendocrinological or
morphological aspects of anorexia nervosa or might be the expression of altered neurotransmission following deficient nutritional state. As some differences exists in relative values in
underweight patients and tend to persist in weight gain states, this could support a potential
abnormal cerebral functioning, a different reaction to starvation within several regions of the
brain or different restoration rates according to the region.
Key Words: Anorexia nervosa, PET scan, FDG, cerebral metabolism
BIOL PSYCHIATRY 1996;40:761-768
Introduction
Anorexia nervosa is a complex syndrome where the most
prominent symptom is weight loss. Different clinical
characteristics, however, such as disturbances in the perFrom the Department of Psychiatry, (VD, FL, YS), PET/Biomedical Cyclotron Unit
(SG, AL), Hrpital Erasme and IRIBHN Statistical Unit (VD), Universit6 Libre
de Bruxelles, Brussels, Belgium.
Address reprint requests to Vrronique Delvenne, MD, Department of Psychiatry,
Hrpital Erasme, Universit6 Libre de Bruxelles, 808 route de Lennik, 1070
Brussels, Belgium.
Received June 5, 1995; revised September 14, 1995.
© 1996 Society of Biological Psychiatry
ception of body shape, high levels of motor restlessness
and hyperactivity are associated with strategies to induce
weight loss such as restricting food, vomiting, purging, or
abuse of diuretics or appetite suppressants.
A number of metabolic and endocrine abnormalities,
mainly at the hypothalamic level, have been described in
anorexia nervosa (Wakeling 1985). Much evidence suggests that most of these changes are secondary to low
weight and caloric deprivation, as they revert to normal
with weight restoration and eating a normal diet.
0006-3223/96/$15.00
SSDI 0006-3223(95)00522-6
762
BIOL PSYCHIATRY
1996;40:761-768
Concerning central nervous system neurotransmitters, a
significant reduction of cerebrospinal fluid (CSF), metabolites of serotonine and dopamine (Kaye et al 1984), or
opiate-related peptides (Kaye et al 1987) have been
observed in anorexia nervosa. These levels do normalize
with weight restoration, however, long-term weight-restored anorectics had elevated concentrations of CSF
5-HIAA (Kaye et al 1991) and a reduction of CSF
norepinephrin levels (Kaye et al 1984). These findings
might be correlated to some clinical characteristics such as
altered mood, anxiety, or obsessional thinking observed in
anorexia nervosa. It has also been postulated that these
neurochemical changes precede the onset of abnormal
eating behavior and are the consequence of a primary
stress reaction (Donohoe 1984).
A possible genetic vulnerability to environmental stress
has been suggested following family and twin studies
(Holland et al 1988). What is not known is the nature and
the importance of this vulnerability in the etiology and in
the development of anorexia nervosa.
Structural brain changes in anorexia nervosa, characterized by a reversible decrease of brain volume, were
described in a number of CT scan studies. These ventricular and sulcal enlargements seem related to malnourishment and are partly reversible after weight restoration
(Artmann et al 1985; Krieg et al 1988), however no
satisfactory explanation has been proposed for these morphological brain alterations.
Supporting the hypothesis of a central nervous system
disorder, neuropsychological evaluation of patients with
anorexia nervosa showed impaired cognitive performance
which appeared to be a sign of poor prognosis at follow-up
(Fox 1981; Hamsher et al 1981; Small et al 1983).
Brain glucose metabolism evaluated by Positron Emission Tomography (PET) is a dynamic method to evaluate
cerebral functioning. In low-weight anorectic patients we
have found an absolute global and regional hypometabolism of glucose which is most important in frontal and
parietal cortices (Delvenne et al 1995). In order to evaluate
the reversibility of this cerebral dysfunction, we evaluated
glucose cerebral metabolic rates (CMRglu) in 10 anorectic
patients in their underweight state and after weight gain.
Ten sex and age-matched subjects were used as controls.
Methods
Subjects
Ten females meeting DSM-IV criteria for anorexia nervosa (American Psychiatric Association 1994), who were
hospitalized in the Department of Psychiatry at Erasme
Hospital, Free University of Brussels, consented to participate in this PET study. The patients were studied first in
V. Delvenne et al
their underweight state and then, after weight-gain. Five of
these patients have also participated in our previous study
(Delvenne et al 1995). Their mean age (__+SD) was 20.0 _
3.9 years (range 14-29 years). They all underwent a
structural interview with the Schedule for Affective Disorders and Schizophrenia (SADS, Spitzer et al 1978) for
patients above 16 years and the Schedule for Affective
Disorders and Schizophrenia for School Age Children
(K-SADS, Chambers et al 1985) for patients younger than
16. Six patients were restrictive anorectics while the
remaining four displayed episodes of binge-eating. Major
depressive disorder was present in two patients while the
others were free of any comorbidity. All patients were free
of psychotropic medication for at least 10 days and none
had ever received neuroleptics or electroconvulsive therapy. Apart from emaciation and secondary amenorrhea,
underweight anorectic patients had no significant abnormalities on physical and neurological examination including electroencephalogram. Laboratory screening tests
were within normal range including serum protein level.
Concerning thyroid functioning, three patients had a low
T3. For the underweight anorectic patients, PET studies
were performed following the first week of voluntary
hospitalization. At that time, patients ate regular hospital
meals with caloric intake around 1500-1800 kcal and they
were controlled for binge-eating and purging. Before PET
examination, patients fasted for two hours to allow stable
plasma glucose levels during the PET procedure. At that
time, starvation ketosis was excluded by means of serum
glucose level which was in the normal range (70-100
mg/dl) or by a negative test for ketones in urine. The
anorectic patients were re-examined after weight gain
following a period of behavior inpatient therapy.
The Body Mass Index (BMI, Rolland-Cachera 1985,
1988), which is the weight over the squared height ratio,
was used as measure of corpulence. Underweight anorectic girls were under 85% of normal weight (mean weight
_ SD in kilograms = 37.3 ___ 5.1; mean height ± SD in
meters = 1.64 + 0.1; mean BMI ___ SD = 13.7 _ 1.7). At
the time of the second PET examination, anorectic patients
had an average weight gain of 20% (mean weight - SD in
kilograms = 48.8 ___5.1; same height; mean BMI --- SD =
18.1 _+ 1.7).
Data was compared with those from ten healthy and
age-matched females (mean age _ SD of 23.8 __+ 4.8
years, ranging from 18-30 years). They were screened for
a negative personal and family history of psychiatric
illness and they had no medical problems. All of them had
a normal corpulence (mean weight -+ SD in kilograms =
59.8 -4- 10.8; mean height _ SD in meters = 1.65 --+ 0.1;
mean BMI ___ SD = 21.9 _ 3.3).
Patients and controls displayed the same educational
and socio-economic status. The patients, and when neces-
PET Scan Study in Anorexia
sary, their parents gave informed consent to participate in
this study. Use of PET with 18 F-FDG for the study of
mental disorders has been approved by the ethics committee of our hospital.
Ratings
Evaluation was completed on the day of the PET study by:
1. The Eating Attitude Test (Garner and Garfinkel
1979): mean score during the underweight state _+
SD of 56.6 _+ 22.8.
2. The 24-item Hamilton Depression Rating Scale
(Hamilton 1960): mean score during the underweight state _ SD of 19.4 _+ 11.9 (range 4-40).
3. The Edinburgh Handedness Inventory (Oldfield
1971): one anorectic patient and one control girl
were left-handed, the remaining ones were righthanded.
Immediately after the PET procedure, patients and control
subjects filled in the 20-item Spielberger State Anxiety
Questionnaire (1968):
scale I and II scores for the underweight
anorectics _+ SD: 47.1 + 10.5 and 58.1 + 12.3.
--Mean
scale I and II scores for the weight-restored
anorectics + SD: 42.2 _+ 7.9 and 52.7 ___ 7.4.
- - Mean scale I and II scores for control subjects + SD:
28.5 + 4.0 and 35.6 -+ 4.8.
--Mean
Experimental Procedure
Intravenous bolus of three to five mCi of FDG was
administered to each subject resting in a supine state with
eyes closed and ears unplugged, in a quiet and dimly
lighted room. Fifteen seconds to 50 minutes after intravenous injection, 21 serial blood samples were centrifuged
and plasma was counted in a gamma counter. Glucose
level was determined in four blood samples at 0, 10, 20,
and 40 minutes and the mean value was used to calculate
cerebral metabolic rates of glucose (CMRglu).
The activity of the tracer in the brain was measured with
an 8-ring CTI-Siemens tomograph (933-08-12. Knoxville,
TN) providing fifteen 7 ram-thick adjacent slices through
the entire brain. FDG was automatically synthesized using
the procedure described by Hamacher et al (1986).
After a 10-minute blank scan acquisition, the subjects
were positioned in the PET camera so as to obtain slices
parallel to the canthomeatal. Subject immobility was
checked using crossed laser beams and lines drawn on the
subject's face. Using a source filled with a 18F-fluoride
solution, a 10-minute transmission scan was performed.
Forty minutes after the FDG injection, a 20-minute emis-
BIOLPSYCHIATRY
1996;40:761-768
763
sion scan was started. Plasma and brain radioactivity were
corrected for decay and PET images were corrected for
attenuation using transmission scan data.
Glucose computation was performed using the CTI
software base on the model of Sokoloff et al (1977)
adapted by Phelps et al (1979). The value used for lumped
constant was 0.42. Values were expressed in p~mol/100g/
min.
To evaluate regional cerebral glucose metabolic rates
(rCMRglu), we delineated 47 irregular regions of interest
(ROI), 20 bilateral, and four medial regions, on PET slices
following a template base on the human brain stereotaxic
atlas of Talairach (1988). All ROI were grouped in eight
anatomical regions (Delvenne et al 1995). Thalamus,
candate nucleus, and putamen, appearing on several adjacent slices, were studied as a surface-weighted mean of
CMRglu: sum of the mean values of glucose (rn0 multiplied by the number of pixel (Pi) and divided by the
number of pixel = 1~ (m i X pi)/'Zpi. If a structure was not
clearly visualized at one of the predefined levels, the
average was calculated for this patient on a reduced
number of slices.
Global absolute cortical values of glucose (ROIc) were
obtained by the mean cortical values which were obtained
by the sum of the mean values of glucose (mi) multiplied
by the number of pixel (Pi) and divided by the number of
pixel = ~(m i • pi)/'Zpi. Left, right, and median cortical
values of glucose (ROI L, ROI R, ROI M) were obtained
similarly by the mean homolateral or median cortical
values.
Relative regional metabolic rates of glucose (rCMRglure 0 were calculated as absolute ROI values divided by
these cortical values of glucose (ROI L, ROI R, ROIM)
respectively for left, right, and median ROI.
Statistical Analysis
In a first analysis ROI 6, ROI L, ROI R, and ROI M rCMRglu
were used as estimates of global, left and right, and
median hemispheric glucose metabolism. They were compared between underweight anorectics and controls and
between weight-gained anorectics and controls with the
use of a Student t test.
Eight neuroanatomic regions were analyzed with multivariate analyses of variance (MANOVA--SPSS, 1988):
(1) superior frontal cortex, (2) lateral inferior frontal
cortex, (3) median inferior frontal cortex, (4) parietal
cortex, (5) temporal cortex, (6) thalamus, (7) caudate
nuclei, (8) putamen. The MANOVA included one between
subjects factor at two levels (underweight anorectic patients, controls) and one repeated measures factor at eight
levels (the eight zones). They were performed on the mean
left/right values of glucose. All main effects and
764
B1OLPSYCHIATRY
V. Delvenne et al
1996;40:761-768
Table 1. Absolute Metabolic Rates for Glucose Respectively
in the Global Cortex Then in Left, Right and Median Cortices
(IxMoles of Glucose/Min/100 g of Brain + SD)
Normal
controls
Global cortex
Left cortex
Right cortex
Median
cortex
29.64
29.93
29.73
28.94
± 1.93
± 1.68
± 2.06
_+ 1.81
U W anorectic
patients a
23.23
23.40
23.11
22.78
±
±
±
±
4.93
5.06
5.01
4.42
Table 2. Mean Left-Right Absolute Regional Cerebral
Metabolic Rates for Glucose (ixMoles of Glucose/Mirdl00 g
of Brain _+ SD)
Normal
controls
W - G anorectic
patients b
28.28
28.46
28.24
27.75
__- 2.96
± 2.98
± 3.13
± 2.65
Student t tests:
~Underweight anorectic patients (UW) vs. controls: p -< 0.003.
bWeight-restored anorectic patients (W-G) vs. controls: non significant.
interactions were introduced into the model. Contrasts of
the "difference" type were performed among the eight
zones. The mean values of each zone were compared
between underweight anorectic patients and controls with
the Student's t test. This procedure was applied on
absolute rCMRGlu and then on relative values.
Underweight and weight-gained anorectic patients were
compared for global, left, and right hemispheric metabolism of glucose, as well as for absolute and relative
rCMRglu of the eight zones, with Student's t tests for
paired samples.
Scores on the BMI and scores of anxiety on the
Spielberger scale were correlated, within each group
(underweight and weight-gained anorectic patients and
controls), with absolute and relative CMRglu values using
the Spearman rank correlation coefficient (Siegel 1956).
Similar correlations were performed for the Hamilton
rating scale for depression but in the underweight anorectic group only.
Superior-frontal cortex
Inferior-frontal cortex
lateral
Inferior-frontal cortex
median
T e m p o r a l cortex
Parietal cortex
Thalamus
Putamen
Caudate
U W anorectic W - G anorectic
patients a
patients b
32.90 _+ 2.71 24.91 --- 5.67 C 30.34 ± 3.32
28.55 ± 2.03 23.21 ± 4.87 d 28.23 ± 3.13
28.13 ± 1.39 22.47 ± 4.49 a 27.25 ± 2.50
23.81
32.23
30.09
33.43
30.56
±
±
±
±
±
1.68
2.50
2.89
2.90
2.94
19.42
23.32
24.71
27.10
25.08
±
±±
±
±
4.06 a
5.77 C
4.61 d
4.54 d
4.11 d
23.03
29.41
29.09
31.99
28.93
~ 2.59
__. 3.39 ~
± 3.46
± 3.65
__. 3.78
aGlobal comparison between underweight anorectic patients (UW) and normal
controls (MANOVA): Group difference: p = 0.001; region difference: p < 0.001;
the interaction was not significant.
bGlobal comparison between weight-gained anorectic patients (W-G) and
normal controls (MANOVA): Group difference: p = 0.9; region difference: p <
0.001; the interaction was not significant.
Cp= 0.001.
,lp < 0.001 (within region group differences).
ep < 0.05 (non-significant with Bonferroni's correction) (within region group
differences).
CMRglu of weight-gained anorectic patients was not
statistically different from normal controls in the same
ROI.
Age of patients and controls was not statistically different
(t test, p = 0.07). Underweight and weight-gained anorectic patients appeared to be significantly more anxious (p =
0.001) than controls. At time of the second PET-scan,
weight-gained anorectics still remained anxious and their
scores were not significantly different than at the time of
the first examination. Weight-gained patients still presented significant lower weights than control subjects (t
test for the BMI, p = 0.004).
CORTICAL AND SUBCORTICAL REGIONS.
Considering
the eight mean right/left regions, the absolute rCMRglu
was found to differ significantly between underweight
anorectic patients and controls (group difference, p =
0.001; region difference, p < 0.001). Contrast analysis
considering the differences in the mean values of cortical
and subcortical regions (Table 2) showed a significant
hypometabolism of glucose (p < 0.01) in each of the eight
regions being mostly significant (p = 0.001) in the
superior frontal and parietal cortices of underweight anorectics.
Comparison of absolute rCMRglu between weightgained anorectics and controls in the eight mean right/left
regions showed no significant group difference (group
difference, p = 0.9; region difference, p < 0.001).
Nevertheless, when looking at the difference between
weight-gained patients and controls within each of the
eight regions, we observed a lower metabolism which
persisted in the parietal cortex (t9 < 0.05) at the time of the
second evaluation.
Absolute Glucose Metabolic Rates
Relative Glucose Metabolic Rates
GLOBAL CORTICAL METABOLISM.
Absolute CMRglu
in the global cortex (ROI G) as well as in ROI L, ROI R, and
ROI M (Table 1) was significantly different between underweight anorectic patients and controls (p -< 0.003).
Only a slight difference between underweight anorectic
patients and controls was found after analysis of the
relative rCMRGlu considering globally the eight mean
right/left regions (group difference, p = 0.057; region
Results
Clinical Data
PET Scan Study in Anorexia
BIOLPSYCHIATRY
765
1996;40:761-768
Table 3. Mean Left-Right Relative Regional Cerebral
Metabolic Rates for Glucose
Normal
controls
Superior-frontal cortex
Inferior-frontal cortex
lateral
Inferior-frontal cortex
median
T e m p o r a l cortex
Parietal cortex
Thalamus
Putamen
Caudate
U W anorectic W - G anorectic
patients a
patients b
1.11 4- 0.04
0.95 ± 0.05
1.07 _+ 0.03 a
1.00 - 0.04 a
1.07 -+ 0.03
1.00 _+ 0.03 e
0.97 _+ 0.02
0.98 -_+ 0.02
0.98 ± 0.02
0.81
1.09
1.02
1.14
1.03
0.84
1.02
1.07
1.18
1.09
0.81
1.03
1.04
1.13
1.03
± 0.05
_+ 0.06
_+ 0.07
_+ 0.05
_+ 0.06
_+ 0.02
--_ 0.05 c
± 0.07
± 0.07
± 0.06 a
_+ 0.03
__+0.04 e
_+ 0.06
± 0.05
--_ 0.07
aGlobal comparison between underweight anorectic patients (UW) and normal
controls. (MANOVA): Group difference: p = 0.057; region difference: p < 0.00l;
the interaction was not significant.
bGlobal comparison between weight-gained anorectic patients (W-G) and
normal controls. (MAN'OVA): Group difference: p = 0.2; region difference: p <
0.001; the interaction was not significant.
Cp = 0.01.
dp < 0.05 (within region group difference).
ep < 0.05 (non-significant with Bonferroni's correction) (within region group
difference).
difference, p < 0.001). Contrasts analysis of the "difference" type was performed on the mean relative values of
cortical and subcortical regions. It showed a significant
hypometabolism in the parietal cortex (p < 0.01), in the
frontal superior cortex (p < 0.05) and a relative hypermetabolism of the caudate nuclei and the inferior frontal
cortex (p < 0.05).
When the eight mean right/left regions were compared
between weight-gained anorectics and controls, no significant difference appeared (group difference, p = 0.2;
region difference, p < 0.001). Contrasts analysis, nevertheless, showed a relative parietal hypometabolism (p <
0.02) and a relative inferior frontal hypermetabolism
(Table 3).
Correlations
In the group of underweight anorectics, as well as in the
weight-gained or control group, no correlation could be
found between absolute rCMRGlu and BMI in each
regions. Similarly, no significant correlation was noted
between anxiety scores and absolute rCMRGlu (eight
regions) within groups (patients and controls). No correlation was found in the underweight anorectic group
between the Hamilton depression score and rCMRGlu.
Comment
This study confirms that underweight anorectic patients
have global and regional absolute glucose cerebral hypometabolism especially important in the frontal and parietal
cortices (Delvenne et al 1995). In relative values, while no
global difference was observed between underweight anorectic patients and controls, comparisons within regions,
nevertheless showed, a hypometabolism in the parietal and
superior frontal cortices associated with a relative hypermetabolism in the caudate nuclei and in the inferior frontal
cortex. After weight gain, following cognitive therapy and
hospitalization, the absolute hypometabolism increased in
all regions while the relative metabolism of glucose tended
to remain abnormal in the parietal and inferior frontal
cortices.
Low body weight or prolonged starvation state can be
proposed to explain the absolute brain glucose hypometabolism as normalization was observed after weight gain,
however, no correlation was observed between rCMRGIu
and BMI within groups (underweight, weight-gained anorectics, controls). It has been suggested that acido-ketosis
and high plasma levels of ketone bodies could explain
hypometabolism of glucose selectively at the cortical level
(Hawkins et al 1979; Herholz et al 1987) although the
pathophysiological mechanism of this hypometabolism is
unknown. At the time of the first PET, our patients were in
a stable metabolic state controlled by the absence of
ketone bodies in urine and constant normal plasma glucose
levels, however, chronic effect of acido-ketosis on the
brain following chronic starvation periods cannot be totally excluded. At the time of reevaluation, patients received stabilized caloric intake but for most of them, they
were still gaining weight. So, we can not exclude that this
might contribute to the persistent abnormalities.
Alterations in function related to glucose utilization are
a result of changes occurring predominantly in nerve
terminals rather than cell bodies (Roland 1993). Some
central neurotransmitters systems have proved to be altered in anorectic patients with significant reductions of
cerebrospinal fluid (CSF) 5-hydroxyindole-acetic acid,
norepinephrine and MHPG or homovanillic acid levels in
the underweight anorectics (Fava et al 1989). These levels
normalized with weight restoration with elevated concentrations of CSF 5-HIAA in long-term weight restored
anorectics (Kaye et al 1991). Abnormalities of these
modulatory amine transmitter systems could contribute to
the absolute global hypometabolism but not to the relative
regional glucose hypometabolism. Indeed, serotoninergic
and noradrenergic fibers distribute diffusely in all cortical
regions while dopaminergic fibers distribute only in certain cortical regions like the prefrontal cortex, the cingulate cortex, and the archicortex. In addition, it is important
to note that, besides this cerebral glucose hypometabolism,
anorectic patients were also characterized by hyperactivity
and they showed no global cerebral dysfunctioning on
cognitive evaluation.
Abnormal hypothalamic functioning, characterized by
neuroendocrine dysfunctions such as amenorrhea, was
766
BIOLPSYCHIATRY
V. Delvenne et al
1996;40:761-768
observed in our group of underweight anorectics. As some
hypothalamic areas project to the subcortical as well as to
the cortical regions (Saper 1990), this may also be proposed to explain brain glucose hypometabolism, however,
these neuroendocrine dysfunctions have frequently delayed recovery after weight gain and our weight-gained
anorectic patients, who had a normalized rCMRglu, were
still amenorrheic.
Correlations between brain glucose hypometabolism
and CT morphological abnormalities in anorexia nervosa
remains questionable. Enlargement of the external cerebrospinal fluid spaces observed in severely emaciated
anorectic patients in CT scan studies are partly reversed
with weight recovery. The significance of this CT changes
are unclear but different pathogenic mechanisms have
been proposed: an inhibition of brain protein biosynthesis
(Artmann et al 1985), hypercortisolemia causing cerebral
dehydration or low plasma level of triiodothyronine (Krieg
et al 1988), or a neuronal brain damage in those cases
which did not reverse with weight recovery (Martin 1958;
Artmann et al 1985). In this study, CT-scans were not
performed in our patients for practical reasons and to
avoid multiplication of X-ray exams.
Clinical characteristics of the anorectic patients, such as
restrictive or binge-eating or purging types, might be
proposed to explain metabolic changes. These subtypes
did not seem to account for statistical differences while
comparisons of rCMRglu between subgroups of anorectic
patients was not significant, however, the number of
patients for comparisons was reduced. Different psychopathological disturbances such as depression or anxiety
have been demonstrated to influence cerebral glucose
metabolism in different ways. Only two of our underweight anorectic patients had an additional diagnosis of
major depressive disorder which was in remission at the
time of reevaluation. Cerebral glucose metabolism of these
two depressed anorectics did not differ from that of
nondepressed patients. In addition, no correlation was
found between reduced rCMRGlu and the Hamilton score
within the group of underweight anorectics. As our patients had higher anxiety scores than control subjects in
their underweight as in their weight-restored states, anxiety does not seem to influence global and regional hypometabolism of glucose. In addition, these anxiety scores
were not correlated with rCMRGlu within groups.
In relative values, a hypometabolism in the parietal and
superior frontal cortices associated with a relative hypermetabolism in the caudate nuclei and in the inferior frontal
cortex was observed and it tended to remain abnormal
after weight gain particularly in the parietal and inferior
frontal cortices. The relative hypermetabolism in the
caudate nuclei which normalized after weight gain has
also been observed by Herholz et al (1987). This was not
observed in our previous study probably because of a
different way of normalization. This relative hypermetabolism in the caudate nuclei and in the inferior frontal
cortex questions relationships between anorexia nervosa
and obsessive compulsive disorder. Indeed, obsessional
features focused on eating patterns were common in
anorexia nervosa (Rothenberg 1986; Holden 1990) and a
relative hypermetabolism of the orbito frontal gyri, corresponding to our inferior frontal cortex, and of the head of
caudate nuclei has also been described in obsessivecompulsive patients (Baxter et al 1987; Nordahl et al
1989). Persistence of these abnormalities after weight gain
might be associated with the persistence of obsessions
concerning food, weight, and body shape in weightrestored anorectic patients.
A trend towards the persistence of an absolute or
relative glucose parietal hypometabolism is observed in
weight-gained anorectics. It might be related to weight
while not all of our patients have recovered a normal
weight at the time of reevaluation, however, no differences
appeared for absolute or relative rCMRglu of patients with
a BMI of or above 19 at the time of the second PET-scan.
This specific parietal hypometabolism might support a
possible primary cerebral dysfunction or a particular
regional sensitivity to consequences of nutritional deficiencies with different restoration rates according to the
region. Cognitive studies showing distortions of body
image (Bowden et al 1989; Home et al 1991) associated
with the impairment of the awareness of emaciation and
particular arithmetic low performances (Fox 1981; Hamsher et al 1981; Small et al 1983) tend to support a parietal
dysfunction in anorexia nervosa. In addition an elevated
pain threshold was observed in eating disorder patients
(Lautenbacher et al 1990) which was not the consequence
of a peripheral polyneuropathy (Pauls et al 1991). This
might correspond to the parietal hypometabolism observed
in our study, however, separating cause-and-effect relationships is particularly difficult in such observations.
Conclusion
In the present study using PET and FDG, we found global
and regional brain glucose hypometabolism in underweight patients with anorexia nervosa. Pathophysiological
mechanism of this observation is unknown. It might derive
from alterations in neurotransmitters system or from neuroendocrinological or morphological characteristics of
anorexia nervosa possibly associated with low body
weight or nutritional factors. Weight gain was followed by
a normalization of this absolute hypometabolism while a
trend towards a still relative parietal hypometabolism and
inferior frontal hypermetabolism was observed. Further
PET Scan Study in Anorexia
imaging studies comparing anorectic patients with underweight non-anorectic patients or other groups of psychi-
BIOL PSYCHIATRY
1996;40:761-768
767
atric patients should clarify the relationship between glucose hypometabolism and persistent starvation states.
References
American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed, Washington DC:
American Psychiatric Association.
Artmann H, Grau H, Adelmann M, Schleiffer R (1985): Reversible and non-reversible enlargement of cerebrospinal fluid
spaces in anorexia nervosa. Neuroradiology 27:304-312.
Baxter LR, Phelps ME, Mazziotta JC, Guze BH, Swartz JM,
Selin CE (1987): Local cerebral glucose metabolic rates in
obsessive compulsive disorder. Arch Gen Psychiatry 44:211218.
Bowden PK, Touyz SW, Rodriguez PJ, Hensley R, Beumont
PJV (1989): Distorting patient or distorting instrument? Br J
Psychiatry 155:196-201.
Chambers WJ, Puig-Antich J, Hirsch M, Paez P, Ambrosini PJ,
Tabrizi MA, Davies M (1985): The assessment of affective
disorders in children and adolescents by semistructured interview. Arch Gen Psychiatry 42:696-702.
Delvenne V, Lotstra F, Goldman S, Biver F, De Maertelaer V,
Appelboom-Fondu J, Schoutens A, Bidaut LM, Luxen A,
Mendlewicz J (1995): Brain hypometabolism of glucose in
anorexia nervosa: A PET-scan study. Biol Psychiatry 37:161169.
Donohoe TP (1984): Stress-induced anorexia nervosa: implications for anorexia nervosa. Life Sci 34:203-218.
Fava M, Copeland PN, Schweiger U, Herzog DB (1989):
Neurochemical abnormalities of anorexia nervosa and bulimia nervosa. Am J Psychiatry 146:963-971.
Fox CF (1981): Neuropsychological correlates of anorexia nervosa. Int J Psychiat Med 11:285-290.
Garner DM, Garfinkel PE (1979): The Eating Attitudes Test: An
index of the symptoms of anorexia nervosa. Psychol Med
9:273-279.
Hamacher K, Coenen HH, Stocklen G (1986): Efficient stereospecific synthesis of no-carrier-added 2-18 F-fluoro-2deoxy-D-glucose using aminopolyether supported nucleophilic substitution. J Nucl Med 27:235-238.
Hamilton M (1960): A rating scale for depression. J Neurol
Neurosurg Psychiatry 23:56-62.
Hamsher KS, Halmi KA, Benton AL (1981): Prediction of
outcome of anorexia nervosa from neuropsychological status.
Psychiatry Res 4:79-88.
Hawkins RA, Biebuyck JF (1979): Ketone bodies are selectively
used by individual brain regions. Science 205:325-327.
Herholz K, Krieg JC, Emrich HM, Pawlik G, Beil C, Pirke KM,
Pahl JJ, Wagner R, Wienhard K, Ploog D, Heiss W-D (1987):
Regional cerebral glucose metabolism in anorexia nervosa
measured by Positron Emission Tomography. Biol Psychiatry
22:43-51.
Holden NL (1990): Is anorexia nervosa an obsessive-compulsive
disorder? Br J Psychiatry 157:1-5.
Holland AJ, Sicotte N, Treasure J (1988): Anorexia nervosa:
evidence for a genetic basis. J Psychosom Res 32:561-571.
Home RL, Van Vactor JC, Emerson S (1991): Disturbed body
image in patients with eating disorders. Am J Psychiatry
148:211-215.
Kaye WH, Berrettini WH, Gwirtsman HE, Chretien M, Gold
PW, George DT, Jimerson DC, Ebert M (1987): Reduced
cerebrospinal fluid levels of immuno-reactive propiomelanocortin released peptides (including Beta-endorphin) in anorexia nervosa. Life Sci 41:2147-2155.
Kaye WH, Ebert MH, Raleigh M, Lake CR (1984): Abnormalities in CNS monoamine metabolism in anorexia nervosa.
Arch Gen Psychiatry 41:350-355.
Kaye WH, Gwirtsman HE, George DT, Ebert MH (1991):
Altered serotonin activity in anorexia nervosa after long-term
weight restoration. Arch Gen Psychiatry 48:556-562.
Krieg JC, Pirke KM, Lauer C, Backmund H (1988): Endocrine,
metabolic, and cranial computed tomographic findings in
anorexia nervosa. Biol Psychiatry 23:377-387.
Lautenbacher S, Michaels AP, Strian F, Pirke KM, Krieg JC
(1990): Pain perception in patients with eating disorders.
Psychosom Med 52:673-682.
Martin F (1958): Pathologie des aspects neurologiques et psychiatriques de quelques manifestations carentielles avec troubles digestifs et neuro-endocriniens. Etude des alttrations du
syst~me nerveux central dans deux cas d'anorexia survenue
chez la jeune fille (dite anorexic mentale). Acta Neurol Belg
58:816-830.
Nordahl TE, Benkelfat C, Semple WE, Gross M, King AC,
Cohen RM (1989): Cerebral glucose metabolic rates in
obsessive compulsive disorder. Neuropsychopharmacology
2:23-28.
Oldfield RC (1971): The assessment and analysis of handedness:
The Edinburgh Inventory. Neuropsychologia 9:97-113.
Pauls AM, Lautenbacher S, Strian F, Pirke KM, Krieg JC (1991):
Assessment of somatosensory indicators of polyneuropathy in
patients with eating disorders. Eur Arch Psychiatry Clin
Neurosci 241:8-12.
Phelps ME, Huang SC, Hoffman EJ, Selin C, Sokoloff L, Kuhl
DE (1979): Tomographic measurement of local cerebral
glucose metabolic rate in humans with (F-18) 2-Fluoro-2Deoxy-D-Glucose: validation of method. Ann Neurol 6:371388.
Roland PE (1993): Cellular metabolism and blood flow. In Brain
Activation, New-York: Wiley-Liss, pp 51-82.
Rolland-Cachera MF (1985): BMI Charts (1 month-21 years).
Graphique S.C.G. France, Institut National de la Sant6 et de
la Recherche Mtdicale (INSERM).
Rolland-Cachera MF (1988): BMI Charts (I month-85 years
and over). France, Institut National de la Sant6 et de la
Recherche Mtdicale (INSERM).
Rothenberg A (1986): Eating disorder as a modern obsessivecompulsive syndrome. Psychiatry 49:45-53.
Saper CB (1990): Hypothalamus. In Paxinas G The Human
Nervous System, San Diego: Academic Press, pp 389-413.
768
BIOL PSYCHIATRY
V. Delvenne et al
1996;40:761-768
Siegel S (1956): Non parametric statistics for the behavioral
sciences. McGraw-Hill Kogakusha, LTD.
Small A, Madero J, Teagno L, Ebert M (1983): Intellect,
perceptual characteristics, and weight gain in anorexia nervosa. J Clin Psychology 39:780-782.
Sokoloff L, Reivich M, Kennedy C (1977): The (14 c)-deoxyglucose method for the measurement of local cerebral glucose
utilization: Theory, procedure and normal values in the
conscious and anesthetized albino rat. J Neurochem 28:897916.
Spielberger 20-item State Anxiety Questionnaire-Manual (1968):
Palo Alto, California: Consulting Psychologists Press.
Spitzer RL, Endicott J (1978): Schedule for Affective Disorders
and Schizophrenia. New York: New York State Psychiatric
Institute.
SPSS/PC+ V2.0 (1988): SPSS Inc; Chicago.
Talairach J, Tournoux P (1988): Co-planar stereotaxic atlas of
the human brain-3-Dimensional Proportional System: An
approach to cerebral imaging. Georg Thieme Verlag, New
York: Thieme Medical Publishers, Inc.
Wakeling A (1985): Neurobiological aspects of feeding disorders. J Psychiat Res 19:191-201.